Summarizing vehicle driving decision-making methods on vulnerable road user collision avoidance

Yuan, Quan; Gao, Yiwei; Zhu, Jiangqi; Xiong, Hui; Xu, Qing; Wang, Jianqiang · 2023 · OpenAlex-citations

DOI: 10.48130/dts-2023-0003

archive: archived pipeline: cataloged verified

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This review paper addresses the critical safety challenges posed by vulnerable road users (VRUs), such as pedestrians and cyclists, in the context of intelligent and autonomous vehicle development. Motivated by high accident rates and significant property losses involving VRUs, particularly in complex mixed-traffic environments like China, the study aims to summarize current research on vehicle driving decision-making methods for collision avoidance. The authors identify a shift in research focus from traditional accident cause analysis to emerging technologies like deep learning and virtual reality, while highlighting gaps in current active safety systems regarding multi-target perception and behavioral cognition. The study employs a knowledge mapping methodology using bibliometric software CiteSpace and VOSviewer to analyze literature from the Web of Science Core Collection database. The researchers retrieved 846 relevant papers published between 2000 and 2022, using keywords related to VRUs, perception, cognition, and decision-making. The analysis involved co-citation, co-occurrence, and keyword burstness assessments to map research trends, international collaborations, and institutional networks. The review categorizes findings into three core dimensions: environmental perception, behavioral cognition, and collision avoidance decision-making. In terms of perception, the paper reviews algorithms for detecting pedestrians and cyclists, noting a transition from traditional manual feature methods (e.g., HOG, SVM) to deep learning approaches (e.g., Faster R-CNN, YOLO, SSD). While deep learning improves accuracy, challenges remain regarding real-time performance, missed detections in complex backgrounds, and classification confusion between pedestrians and cyclists. Regarding behavioral cognition, the study examines models for intention identification and trajectory prediction, including micro-simulation models (e.g., Social Force Model) and individual behavior models (e.g., Markov chains, SVM). It highlights that current models often rely on hypotheses rather than actual data and struggle with the random and uncertain nature of VRU behavior. In decision-making, the authors note that existing risk assessment models frequently fail to incorporate VRU intention identification and trajectory prediction, with a specific lack of exploration regarding collision risks for cyclists. The significance of this work lies in its comprehensive mapping of the current state of VRU collision avoidance research and its identification of key deficiencies. The authors conclude that accurate identification of VRUs in complex environments and the coupling of intention identification with trajectory prediction are major demands for future research. They emphasize the need for improved joint detection frameworks and more practical behavioral models that bridge the gap between theoretical hypotheses and real-world data. This review provides guidance for enhancing traffic safety for VRUs under intelligent and connected transportation conditions, suggesting that future efforts should focus on integrating perception, cognition, and decision-making to address the unique behavioral attributes of VRUs in mixed traffic environments.

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.

Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

• vru crash typology
• vru facing ehmi